Another important feature of Opteron is that memory bandwidth scales with additional CPUs in an SMP setup.
However, I think the bandwidth aggregates among the local memories attached to each CPU. To get maximum benefit, the OS kernel would need to know how to best schedule and allocate processes to prevent CPUs from accessing other CPUs' memories often.
echo "I think all the current 64-bit CPU producers are thinking about this, too, and saying \"oh\" and \"um\" quite a bit trying to figure out there marketing plans." | sed -e "s/there/their/g"
Gee, Opteron is MUCH less expensive, performs better, runs up to 8-way with off the shelf components and runs your 32-bit x86 code twice as fast and absolutely compatibly. Let me think about this...;-)
I think all the current 64-bit CPU producers are thinking about this, too, and saying "oh" and "um" quite a bit trying to figure out there marketing plans.
It is, because it was originally designed to compete with old "big iron" RISC servers. Itanium is big, it is hot, it is low-volume, and it is expensive.
I would bet that Opteron actually has Intel shitting itself right now. The more I see about Opteron, the better it looks for 1 to 4 CPU servers, which are generally powerful enough for most tasks, now-a-days. Opteron is sitting squarely in competition with Xeon, Itanium, PowerPC, UltraSPARC, MIPS, etc. for a huge segment of the server market. Opteron is going to make this next year very interesting, because 64-bit rack-mount servers with good processing power are now easily under $5,000. I am expecting a shake down in 64-bit pricing across the industry as everyone tries to compete.
Then you'll get to have fun updating libraries whenever you want to install something, as well as patching BIND, sendmail, the kernel, etc.
It doesn't have to be all that bad. Packages are relocatable, so unusually sensitive applications can be put into their own root directory hierarchy. Using NFS wisely can allow for one set of applications on a network (patching once and only once is quite nice). Only one or two servers on the whole network should be running Sendmail and BIND in a vulnerable mode. UNIX is also easier to pare down, so there are much fewer things that need to be patched. With a good network design, patches can be rolled out automatically over SCP, and UNIX machines tend to reboot pretty reliably, unless a patch screws up an init script.
It is just a simple fact that UNIX is less complex than Windows. It has fewer lines of source code, more transparent modularization, strict separation between the GUI and the kernel, widely available and thorough documentation, three decades of experience behind it, almost complete scriptability, among other things. Windows, on the other hand, is as opaque as mud--there could be a golden city under there or just more mud, but we'll never know.
Well, 40,000 is one thing. But "40,000 thousand" is quite another - it's 40 million.
Thanks for clearing that up. I guess writing "40,000 thousand" instead of "40,000,000", to some people, is like turning on auto-hide for the status bar in Windows ("Uh, Mr. Support Person, my compooter is broken, I need you to fix it for me")
Between the costs of patching, the two weeks of downtime per user per year, and the flaws that threaten national security, has no one yet found a good way to sue for damages??? WTF?
The number of patches must be the worst possible metric for measuring bugs. A better measure is: (several bugs per 1000 lines of code) X (40,000 thousand lines of code in Windows) = over 100,000 bugs in Windows. Thus, it follows that (100,000 bugs/installation) X (100,000,000 installations) = 10,000,000,000,000 Windows bugs worldwide.....OMG, the plauge of the apocolypse is upon us!
The difficult question is whether the costs of patching outweigh the costs of NOT patching.
The lowest cost method of patching Windows is to use the special one-time CD-ROM distributions available out there that fix Windows once and for all. I can recommend several brands: Slackware Disk 1, Red Hat Disk 1, Solaris Install Disk 1, and OpenBSD Disk 1. There are other very good ones, as well, but they all have about the same level of Windows-patch elimination power.
How long until "Platinum" electronically enhanced condoms are offered along with the "male enhancement pills"? I can hear them now...Bzzzzzzzzzzzzzzz...bzt..zzt...Ahhhh.
If you can wait a few months to pick up a particular game, the price usually drops to $25 or so, used.
After a year or so, even unused games drop to about US$20. I've never payed full price for a game in the last three years ($29.99 being the highest price I've paid for a slightly newer game). I've found that the local Toys'R'Us has a good selection of old games. Shopping malls are the worst, and flea markets are a mixed experience, depending on the vendor.
And rent games before you buy.
This is often good advice for new games. I rented Star Wars: The Clone Wars and finished the damn thing before it was due back. It feels good to spend only $5 on a game. For old/used games, the rental fee is too great a percentage of the purchase price, so renting is less worthwhile, then.
Perhaps not a dime, but there are literally dozens (if not hundreds) of collaboration-solution-spouting companies pushing their dubious products. Just try searching on Google for a while, and you'll see every flavor of collaboration snake oil imaginable.
Regardless, it appears video conferencing isn't all that bandwidth-intensive. Sun's SunForum prouct claims to require only 25Kbit/sec by default. I've never used it, but that's what their FAQ says. I'm sure having MPEG hardware doesn't hurt, either.
I haven't used NetMeeting, either. Does that work?
CASE tools take those decisions out of your hands.
Or worse, CASE tools put those decisions into the hands of someone who thinks they are now a database architect, because they can pointy-clicky the tables around on a screen. CASE tools cannot infer the meaning of the data, and, therefore, are actually dangerous in the hands of some PRB who thinks "normalization" is something that happens to immigrants.
Re:Sony already did this
on
High Density CDs
·
· Score: 2, Interesting
These little fart in a jar techs will no doubt go the way of the zip drive.
Yeah, I bought a 250MB Zip drive right before the CD-R boom. That was a regrettable purchase, when everyone else was burning twice the capacity for a fraction of the cost. I can't imagine that those newer 750MB Zip drives are even selling the first production run.
My question is this: where are OpenSource design tools?
The best high-level design tools for UML, database schemas, etc. are still proprietary. This is also the reason I hesitate to use such tools in the first place. For example, how does a binary project file get effectively version-controlled? How can it be shared with people who haven't blown $1,500 for SuperDesignToolXYZ?
I've also seen thousands of dollars blown on design tools, when the project is so small, anyway, that plain text files would have sufficed. Or, better, actually documenting the class definitions in a Java or C++ project.
Don't forget that design tools also have their own built-in and distracting learning curve. It is easy to waste time poking around in a tool without actually accomplishing real work. If a company is going to invest in the cost of a design tool, at least send the team to a week of training.
In short, design tools are always a mixed blessing and should be viewed skeptically.
They are free to speak, we are free to not listen or to not pass their messages on.
When "speech" becomes effectively a Denial of Service attack, freedom of speech ends, IMO.
Examples:
SPAM -- literally reducing peoples' ability to communicate effectively. This hurts individuals and businesses. The cost to the recipient is real.
Loud Music -- that bass pumping out of my asshole neighbor's house is not protected speech. It distrupts my family, my quality of life, my own attempts at speech, and is, like SPAM, bad for society.
Grafitti -- it is vandalism and not art.
There are the other classic examples like yelling "fire" when there isn't one. Burning a cross in someone's yard is, also, definitely not free speech.
People who piss on other people's lives using Free Speech as an excuse are among the lowest examples of humanity. They deserve no sympathy.
OpenBSD looks fantasic and I was about to give it a whirl when I realized they don't support SMP.
Consider what OpenBSD excels at and consider these questions:
Does a firewall really need two 2GHz CPUs?
How about a router, modest fileserver, or e-mail server?
Considering the complexity that SMP would probably add to the kernel (race conditions, data integrity, etc.), it may be counter-productive towards the goal of uncompromising security.
For bigger servers (4 or more CPUs) just run Solaris, FreeBSD, or Linux behind OpenBSD-based infrastructure. I think this is a tasty compromise.
Slashdot Mirror
"Unix survives only because everyone else has done so badly." - Donald A. Norman
As true today as when it was written...
(drops head in shame and sighs)
How about the dog house?
Another important feature of Opteron is that memory bandwidth scales with additional CPUs in an SMP setup.
However, I think the bandwidth aggregates among the local memories attached to each CPU. To get maximum benefit, the OS kernel would need to know how to best schedule and allocate processes to prevent CPUs from accessing other CPUs' memories often.
echo "I think all the current 64-bit CPU producers are thinking about this, too, and saying \"oh\" and \"um\" quite a bit trying to figure out there marketing plans." | sed -e "s/there/their/g"
Gee, Opteron is MUCH less expensive, performs better, runs up to 8-way with off the shelf components and runs your 32-bit x86 code twice as fast and absolutely compatibly. Let me think about this... ;-)
I think all the current 64-bit CPU producers are thinking about this, too, and saying "oh" and "um" quite a bit trying to figure out there marketing plans.
...Itanium is so incredibly expensive.
It is, because it was originally designed to compete with old "big iron" RISC servers. Itanium is big, it is hot, it is low-volume, and it is expensive.
I would bet that Opteron actually has Intel shitting itself right now. The more I see about Opteron, the better it looks for 1 to 4 CPU servers, which are generally powerful enough for most tasks, now-a-days. Opteron is sitting squarely in competition with Xeon, Itanium, PowerPC, UltraSPARC, MIPS, etc. for a huge segment of the server market. Opteron is going to make this next year very interesting, because 64-bit rack-mount servers with good processing power are now easily under $5,000. I am expecting a shake down in 64-bit pricing across the industry as everyone tries to compete.
The only meaningful benchmark IMO is processing_power/cost.
:)
So, you would be happy buying only eMachines to put in your server room ?
They use the sled to examine the interactions between weapons and targets in a controlled dynamic environment.
That's what they wrote, but, really, they just thought it would be cool to see something going mach 8 hit a wall. Who wouldn't want to see that?
Q: How does a Unix guru have sex? A: unzip;strip;touch;finger;mount;fsck;more;yes;umoun t;sleep
A real UNIX guru would put that into a script run by a cron task that pages him (obviously a him writing scripts like this) upon successful execution.
What's going to happen, a trekkie is going to lose his virginity?
Soooo...where do trekkies come from in the first place? Are they mutants?
I don't buy that argument (although you may just be joking).
I'm partly joking. Windows is a much higher-risk venture than any UNIX/Linux/BSD system I'm aware of.
Then you'll get to have fun updating libraries whenever you want to install something, as well as patching BIND, sendmail, the kernel, etc.
It doesn't have to be all that bad. Packages are relocatable, so unusually sensitive applications can be put into their own root directory hierarchy. Using NFS wisely can allow for one set of applications on a network (patching once and only once is quite nice). Only one or two servers on the whole network should be running Sendmail and BIND in a vulnerable mode. UNIX is also easier to pare down, so there are much fewer things that need to be patched. With a good network design, patches can be rolled out automatically over SCP, and UNIX machines tend to reboot pretty reliably, unless a patch screws up an init script.
It is just a simple fact that UNIX is less complex than Windows. It has fewer lines of source code, more transparent modularization, strict separation between the GUI and the kernel, widely available and thorough documentation, three decades of experience behind it, almost complete scriptability, among other things. Windows, on the other hand, is as opaque as mud--there could be a golden city under there or just more mud, but we'll never know.
Well, 40,000 is one thing. But "40,000 thousand" is quite another - it's 40 million.
Thanks for clearing that up. I guess writing "40,000 thousand" instead of "40,000,000", to some people, is like turning on auto-hide for the status bar in Windows ("Uh, Mr. Support Person, my compooter is broken, I need you to fix it for me")
Between the costs of patching, the two weeks of downtime per user per year, and the flaws that threaten national security, has no one yet found a good way to sue for damages??? WTF?
(measured in terms of number of patches.)
The number of patches must be the worst possible metric for measuring bugs. A better measure is: (several bugs per 1000 lines of code) X (40,000 thousand lines of code in Windows) = over 100,000 bugs in Windows. Thus, it follows that (100,000 bugs/installation) X (100,000,000 installations) = 10,000,000,000,000 Windows bugs worldwide.....OMG, the plauge of the apocolypse is upon us!
The difficult question is whether the costs of patching outweigh the costs of NOT patching.
The lowest cost method of patching Windows is to use the special one-time CD-ROM distributions available out there that fix Windows once and for all. I can recommend several brands: Slackware Disk 1, Red Hat Disk 1, Solaris Install Disk 1, and OpenBSD Disk 1. There are other very good ones, as well, but they all have about the same level of Windows-patch elimination power.
How long until "Platinum" electronically enhanced condoms are offered along with the "male enhancement pills"? I can hear them now...Bzzzzzzzzzzzzzzz...bzt..zzt...Ahhhh.
If you can wait a few months to pick up a particular game, the price usually drops to $25 or so, used.
After a year or so, even unused games drop to about US$20. I've never payed full price for a game in the last three years ($29.99 being the highest price I've paid for a slightly newer game). I've found that the local Toys'R'Us has a good selection of old games. Shopping malls are the worst, and flea markets are a mixed experience, depending on the vendor.
And rent games before you buy.
This is often good advice for new games. I rented Star Wars: The Clone Wars and finished the damn thing before it was due back. It feels good to spend only $5 on a game. For old/used games, the rental fee is too great a percentage of the purchase price, so renting is less worthwhile, then.
Perhaps not a dime, but there are literally dozens (if not hundreds) of collaboration-solution-spouting companies pushing their dubious products. Just try searching on Google for a while, and you'll see every flavor of collaboration snake oil imaginable.
Regardless, it appears video conferencing isn't all that bandwidth-intensive. Sun's SunForum prouct claims to require only 25Kbit/sec by default. I've never used it, but that's what their FAQ says. I'm sure having MPEG hardware doesn't hurt, either.
I haven't used NetMeeting, either. Does that work?
CASE tools take those decisions out of your hands.
Or worse, CASE tools put those decisions into the hands of someone who thinks they are now a database architect, because they can pointy-clicky the tables around on a screen. CASE tools cannot infer the meaning of the data, and, therefore, are actually dangerous in the hands of some PRB who thinks "normalization" is something that happens to immigrants.
These little fart in a jar techs will no doubt go the way of the zip drive.
Yeah, I bought a 250MB Zip drive right before the CD-R boom. That was a regrettable purchase, when everyone else was burning twice the capacity for a fraction of the cost. I can't imagine that those newer 750MB Zip drives are even selling the first production run.
My question is this: where are OpenSource design tools?
The best high-level design tools for UML, database schemas, etc. are still proprietary. This is also the reason I hesitate to use such tools in the first place. For example, how does a binary project file get effectively version-controlled? How can it be shared with people who haven't blown $1,500 for SuperDesignToolXYZ?
I've also seen thousands of dollars blown on design tools, when the project is so small, anyway, that plain text files would have sufficed. Or, better, actually documenting the class definitions in a Java or C++ project.
Don't forget that design tools also have their own built-in and distracting learning curve. It is easy to waste time poking around in a tool without actually accomplishing real work. If a company is going to invest in the cost of a design tool, at least send the team to a week of training.
In short, design tools are always a mixed blessing and should be viewed skeptically.
They are free to speak, we are free to not listen or to not pass their messages on.
When "speech" becomes effectively a Denial of Service attack, freedom of speech ends, IMO.
Examples:
SPAM -- literally reducing peoples' ability to communicate effectively. This hurts individuals and businesses. The cost to the recipient is real.
Loud Music -- that bass pumping out of my asshole neighbor's house is not protected speech. It distrupts my family, my quality of life, my own attempts at speech, and is, like SPAM, bad for society.
Grafitti -- it is vandalism and not art.
There are the other classic examples like yelling "fire" when there isn't one. Burning a cross in someone's yard is, also, definitely not free speech.
People who piss on other people's lives using Free Speech as an excuse are among the lowest examples of humanity. They deserve no sympathy.
OpenBSD looks fantasic and I was about to give it a whirl when I realized they don't support SMP.
Consider what OpenBSD excels at and consider these questions:
Does a firewall really need two 2GHz CPUs?
How about a router, modest fileserver, or e-mail server?
Considering the complexity that SMP would probably add to the kernel (race conditions, data integrity, etc.), it may be counter-productive towards the goal of uncompromising security.
For bigger servers (4 or more CPUs) just run Solaris, FreeBSD, or Linux behind OpenBSD-based infrastructure. I think this is a tasty compromise.